Metaphor as argument: Rhetorical and epistemic advantages of extended metaphors

Transcription

1 Full reference: Oswald, Steve & Rihs, Alain (submitted). Metaphor as argument: Rhetorical and epistemic advantages of extended metaphors. Argumentation. Metaphor as argument: Rhetorical and epistemic advantages of extended metaphors Steve Oswald & Alain Rihs (Université de Neuchâtel) 1. Introduction Metaphors have long been considered to function as rhetorical devices fulfilling strategic goals in argumentative exchanges. As many rhetorical figures, they are believed to be particularly effective, sometimes more than literal formulations, when it comes to convincing an audience. This paper takes this claim seriously and attempts to examine, from an argumentative perspective informed by cognitive insights about information-processing, the extent to which metaphors and in particular extended metaphors may bring about epistemic effects geared to positively affect message acceptance. The purpose of this paper is to assess the argumentative potential of extended metaphors, 1 in particular when these are used in political discourse to metaphorically represent entities of the world. Specifically, we will contend that multiple instantiations of the conceptual properties of a same metaphorical construal can fulfil an argumentative function: recurring exploitations of a metaphor can indeed be argumentatively articulated so as to converge towards the justification of the proposed metaphorical construal in some sort of confirmational dynamics, so much so that the metaphor may cease to be perceived as one, turning what was at first metaphorically construed into a representation about an actual state of affairs one can believe to be true. By combining this conception of metaphor as argument with recent work on epistemic vigilance in cognitive anthropology and psychology (see Sperber et al. 2010), we will try to characterise the rhetorical advantages offered by the structure of extended metaphors in terms of speaker ethos but also in terms of the epistemic strength of the content of the message. In other words, we will try to cognitively ground the claim that the careful construction of an extended metaphor may positively impact speaker credibility and trustworthiness but also the audience s perception of the truth of the message. We will illustrate these two arguments with the analysis of a pamphlet published on the website of Oskar Freysinger, a member of the National Council of Switzerland (the lower house of the Federal Assembly) and a representative of the conservative right-wing party Union Démocratique du Centre. In this pamphlet, Freysinger weaves the metaphor of the USA as an empire. We will show that the way the extended metaphor is constructed can be described as an argumentative strategy whose outcome is to convince the audience that the USA is actually literally an empire, and therefore that action should be undertaken for protection. 1 As a preliminary note, let us state that we will not claim here that all extended metaphors are necessarily convincing. Rather, we discuss their potential as an argumentative strategy speakers can opt for in discursive contexts where they want to convince an audience; we will show how this can be done, in light of the analysis of a political pamphlet where an extended metaphor is exploited throughout the text to ground a specific standpoint. 1

2 Our account is meant to explain how extended metaphors can constitute powerful argumentative devices. However, we will neither claim that extended metaphors are necessarily argumentative, nor that they cannot fulfil other functions. We focus on contexts in which speakers pursue legitimising goals and our account should therefore be understood as describing and explaining some strategies that can unfold within said contexts. From an epistemological perspective, we adopt Relevance Theory s account of meaning construction (see for instance Sperber & Wilson 1986, 1995 and Carston 2002), which, as far as metaphor is concerned and contrary to a widespread belief, is compatible to a certain extent with the cognitive linguistic account of metaphors (see Wilson 2011, Gibbs & Tendahl 2006). Following Wilson (2011), who highlights interfaces between both theories, even if the notion of conceptual mapping is not essential to either the production or interpretation of metaphors (2011: 210), we do believe that it can be contextually relevant for an addressee to represent such mappings as he processes metaphors, in particular extended and novel metaphors which are not instances of conceptual metaphors such as LOVE IS A JOURNEY, ARGUING IS FIGHTING or TIME IS SPACE, as our analysis will show. 2 In other words, we posit that the hearer might be led to consider in specific contexts which invite him to do so, and particularly when processing extended metaphors, that the metaphor maps (i.e., systematically establishes correspondences between) representations, thereby inviting him to construe one conceptual domain (the socalled target conceptual domain) in terms of the properties of another representation (the socalled source domain). 3 In other words, we consider that extended metaphors can typically make a particular metaphorical mapping relevant, even more so when what is at stake is an explicit assessment of the proposed metaphorical construal. Underlying our study is also the general assumption that an account of how people process and understand language and in the case at hand, metaphorical language is a first step in accounting for the mechanisms governing belief fixation in communicative exchanges. To sum it up in two key-words, we will focus on the relationship between understanding and believing and accordingly try to highlight the importance of comprehension with respect to beliefs (see also Author1 2011). Section 2 reviews some aspects of metaphor processing and motivates the following assumptions: i) extended metaphors may fulfil an argumentative function and can be used to self-legitimate their conceptual mapping, the consequence of which could be a literalisation of the metaphor and ii) they simultaneously positively impact perceived speaker competence. Section 3 illustrates the suggested account with the analysis of an extended metaphor contained in Freysinger s pamphlet. Section 4 summarises the arguments advanced to this end and concludes on the relevance of cognitive insights to argumentation studies and discourse analysis more generally. 2 We adopt the conventional notation by which speakers are referred to as females and addressees as males. 3 This is why we will occasionally use the traditional terminology of Cognitive Linguistics with the following definitions: a conceptual domain is defined as any coherent organization of experience (Kövecses 2002, p.4); the source domain as the conceptual domain from which we draw metaphorical expressions to understand another conceptual domain (ibid.) and the target domain as the conceptual domain that is understood this way (ibid.). The term mapping denotes a set of systematic correspondences between the source and the target in the sense that constituent conceptual elements of B [the source] correspond to constituent elements of A [the target] (ibid., p.7). 2

3 2. Metaphors, meaning, discourse and argumentation 2.1. Metaphor and meaning: costs and benefits There is a great deal of experimental work on the nature of metaphor processing (see Gibbs & Tendahl 2006 for a review). In particular, quite some effort has been devoted to exploring the issue of whether metaphor processing is cognitively costly, and whether the conventional or novel nature of metaphors influences processing effort. Noveck et al. (2001) and Glucksberg (2001), among others, have shown that conventional metaphors are relatively unproblematic in processing terms: their comprehension does not require significantly longer reading times compared to literal equivalents. One could thus suspect that the degree of conventionality of a given metaphor may play a significant role in the derivation of its meaning. Conventional metaphors might be easily accessible mainly because the same encyclopaedic information has been selected over time as they have been repeatedly processed, which could also be interpreted in terms of the familiarity and salience of their conventional meaning (see Giora 1999). In turn, this would suggest that conventional metaphors convey lexicalised meanings. 4 For instance, if the interpretation of (1) which is conventionally interpreted as meaning something like (2) (1) Jeffrey is a clown. (2) Jeffrey is a joker. is not particularly costly in terms of processing effort, this is presumably because being a joker, among other things, is part of the intension of clown ; the metaphor s conventionality is somehow determined by the accessibility to a prototypical relation (here, the information that clowns are supposed to be jokers by definition). In that respect, the way conventional metaphors are processed does not significantly depart from the way literal language is processed. It could be argued that novel metaphors, on the other hand, can differ from conventional ones in two ways: i) the relevant features involved in the representation are not those which are normally selected; ii) the relevant conceptual properties involved in their interpretation may be completely unexpected, which forces the hearer to build up a representation from conceptual properties that have not previously been mobilised together. To illustrate this last point, imagine that (1) is uttered in a context involving a discussion about Jeffrey s shoe size: in such a scenario the hearer is led to reassess the metaphorical potential of clown in order to single out a less salient property such as a clown s unusually big shoe size, which will allow him to infer that Jeffrey has unusually big feet. In this case, metaphorical creativity is defined by the fact that the properties selected in the source domain are not lexically the most salient ones, and therefore not the most accessible within the concept s intension. Arguably, novel metaphors inherit their creativity from the fact that the particular metaphorical operation they call for has seldom, if never, been processed before. 4 Wilson notes that the process of ad hoc concept formation which is taken to characterise metaphor processing is the same for both lexicalised and non-lexicalised metaphors: The adjustment process may be a spontaneous, one-off affair, involving the construction of an ad hoc concept which is used once and then forgotten; or it may be regularly and frequently followed, by a few people or a group, until, over time, the resulting ad hoc concept may stabilise in a community and give rise to an extra lexicalised sense (Sperber & Wilson 1998; Vega Moreno 2007; Wilson & Carston 2007) (Wilson 2011, p. 209). 3

4 Under this view, the cognitive operations involved in the comprehension of both conventional and novel or creative metaphors are of the same kind. What distinguishes them is whether the information sets required for their interpretation are the ones that are usually mobilised or not. This, however, is not sufficient to conclude that conventionality entails ease of processing and novelty the opposite (see Gibbs and Tendahl 2006). This is because metaphors hardly ever have to be interpreted in neutral contexts. In fact, strong contextual constraints (such as thematic constraints for instance) will generate expectations about the type of contents that will be contextually relevant at a given time. These constraints may very well facilitate metaphor processing, even in cases of novel metaphor. To go back to example (1), it seems reasonable to assume that in a conversation about the size of people s feet, (1) will not cause particular interpretative problems and straightforwardly lead to the interpretation that Laszlo has big feet, precisely because in such conversation participants are expected to contribute information that is relevant to the topic. In Gibbs & Tendahl s terms, it will be the context that determines how quickly we can process a metaphorical utterance of whatever kind (2006, p.396). The importance of context in the interpretation of metaphors thus seems to suggest that, more than processing effort, it is the nature of the information sets selected to perform interpretative tasks that determines ease of processing. In our analysis we will take extended metaphors as a case in point and argue that their processing triggers significant cognitive effort on the addressee s behalf if only because many different properties of the same core metaphorical construal need to be cumulatively processed throughout the text. As we adopt a relevance-theoretic perspective on the issue (as per Sperber & Wilson 1986, 1995, Wilson & Sperber 2012 and Carston 2002), we will consider that the cognitive cost of the procedure has to be offset by some sort of benefit; 5 in the case of extended metaphors argumentatively exploited, for instance in political discourse, we suggest that this benefit can amount to the fulfilment of expectations about the argumentative purpose of the metaphor. Relevance Theory (henceforth RT) distinguishes the interpretative procedure metaphors trigger (which is not idiosyncratic) from the effects they produce (which show some particularities, especially when the metaphor is novel). On the processing side, and similarly to literal approximations, the interpretation of metaphors is taken to rely on a lexical loosening device, which widens the word s literal denotation in order to make for the designation of a neighbouring concept. This mechanism, along with its narrowing counterpart in cases of meaning specification, is referred to as ad hoc concept formation (see e.g. Carston 2002, 2010, Wilson 2003). On this view, lexical meaning including metaphorical meaning is said to accommodate contextual constraints of relevance so that the specific intended meaning communicated by the metaphor is able to vary alongside contextual variation. As Carston puts it, the description of such concepts as ad hoc reflects the fact that they are not linguistically given, but are constructed online (on the fly) in response to specific expectations of relevance raised in specific contexts. (Carston 2002, p. 322). 5 Relevance theory postulates that human cognition is governed by a principle of relevance according to which information is processed following a cost/benefit dynamics. One of the fundamental claims of the account is that establishing the relevance of any verbal input consumes cognitive resources and yields cognitive benefits such as the improvement of one s knowledge of the world (see Sperber & Wilson 1995). Relevance is therefore defined here in terms of balance between processing effort and cognitive effect. 4

5 On this account, ad hoc concepts are taken to be constructed in order to satisfy expectations of relevance in the meaning derivation procedure of lexical items: the lexical meaning of a concept thus becomes a pragmatic function of context, which is determined by the communicative needs of the speaker in every communicative exchange. The first cognitive benefit of the cognitive mechanism of ad hoc concept formation therefore has to be interpreted in terms of its contribution to the identification of speaker meaning. This perspective is in particular the one advocated by Noveck et al. (2001), who evaluate the cognitive benefits of metaphor in terms of better comprehension. Such a benefit is particularly straightforward in the case of conventional metaphors: their comprehension amounts to the derivation of a stable and predictable meaning, and since the latter is conventional, it can be taken to be derived pretty much the same way stable and predictable meanings are derived from literal approximations. The effects produced by novel and creative (or poetic) metaphors in particular can be more complex and somewhat different. The relevance-theoretic model of lexical pragmatics holds in this respect that creative metaphors can be responsible for the generation of an array of weak implicatures. These are implicit contents which express a speaker s complex thought efficiently and faithfully, but which are identified as being intended by the speaker only with a low degree of reliability (see Blakemore 1992, p and Sperber & Wilson 1995, p for a discussion). In particular, the grounds on which addressees identify weak implicatures are far from strong, and as a consequence it may be fairly tricky for the hearer to ascertain whether the speaker intended to make manifest that she wanted the hearer to identify them as being meant by her. In relevance-theoretic parlance, weak implicatures are contents the communicative intention of which is not mutually manifest (see also Sperber & Wilson 2008). A particularly vivid example of this phenomenon is to be found in creative metaphors in poetic texts: the range of weak implicatures they are likely to license is dependent to a great extent on the addressee s own experience of the text he is reading (and obviously on his general background knowledge, preferences, etc.). Example (3) provides a clear illustration of this: (3) My neighbour is a dragon (Blakemore 1992, p.163) When uttering (3), the speaker may convey that her neighbour is unfriendly and fierce, but additional weak implicatures might also be drawn, such as the speaker s neighbour has the appearance of a dragon or the speaker s neighbour is terrifying beyond imagination. These are not necessarily publicly endorsed by the speaker, but they would indeed offset processing cost, to the extent that the speaker has manifestly precisely not uttered the literal equivalent. Yet, in so doing, she has encouraged the addressee to further process the utterance to discover additional implicit contents. These are contents that justify the speaker s utterance as the best means of representing his thoughts, and it is these implicatures which explain why even this rather standardized example of metaphor cannot be paraphrased without loss (Blakemore 1992, p ). Both conventional and creative metaphors achieve relevance by being understood; some of them are furthermore able to trigger weak implicatures, resulting in the derivation of richer meanings. But these are cognitive effects strictly limited to processes of meaning derivation, and, as such, are matters of illocutionary concern. In this paper, we will consider possible perlocutionary effects of metaphoric creativity, in particular as to what regards belief fixation, by defending the idea that the relevance of an extended metaphor through the satisfaction of expectations of 5

6 relevance that can go beyond comprehension can positively affect the acceptance of a belief derived from the content of the metaphor The discursive nature of extended metaphors Extended metaphors are realised in discourse through the recurring exploitation of the same metaphor at several conceptual levels over a relatively long span of text. Their interpretation, in those cases, can accordingly be seen as an incremental process which gradually enriches the representation as different properties of the source domain successively appear in one form or another throughout the same discourse. A given text or discourse contains an extended metaphor when a metaphorical field extends through an entire discourse (Werth 1994, p. 83). They are generally found in poetic and literary works, since they may express more abstract emotional experiences for which no sui generis language exists (ibid. p. 84). They can nevertheless be used for other purposes, to the extent that they may be used to make the expression more striking (the flowers of rhetoric approach) or to allow the topic to be viewed simultaneously from more than a single perspective (ibid.). In all three cases, however, extended metaphor can be characterised as involving poetic choice. From a cognitive pragmatic perspective, such poetic phenomena may be approached through the notion of weak implicature, as seen above, in order to deal with the effects of extended metaphor in terms of meaning potential. As far as cognitive processing is concerned, and assuming that the search for relevant correspondences between the source and target domains can be at the heart of their interpretative process, extended metaphors appear to be demanding in terms of effort, since understanding them and establishing their relevance will require the addressee to perform, if not unexpected conceptual associations, at least multiple ones. Trivially, extended metaphors involve significant cognitive processing because they extend through discourse. Once they are recognised as such, extended metaphors will thus generate expectations for further and perhaps even full exploitations of the conceptual mapping. What an extended metaphor does, then, is encourage the addressee to keep on exploring several aspects of the proposed metaphor in terms of an elaboration of the construal the speaker wants to convey. Extended metaphors can consequently be said to involve complex multi-stage representational operations triggered cumulatively as discourse unfolds. The processes involved in these representational steps are inferential; that is, they require the hearer to combine information contained in the linguistic form with contextual information in order to derive conclusions in the form of additional, new, representations. From the analysis of an excerpt of A Passage to India by E.M. Forster, Werth suggests that the inferences required to process the extended metaphor can also lead to the revelation of an underground metaphor (Werth 1994, p. 85) which the hearer is supposed to extract from the text. Extended metaphors may therefore provide the grounds for rich inferential work geared towards the derivation of specific conclusions. We owe this possibility to the discursive nature of extended metaphors: the conclusions we draw from them are cumulative, and, crucially, achieved by way of text and discourse processes, rather than sentence processes (ibid.). The discursive realisation of the metaphorical elaboration involved in extended metaphors makes it approachable with discourse-analytical tools. Notably, as already highlighted by several scholars (see e.g., Kimmel 2009 and Koller 2005), the discursive notion of coherence could be 6

7 used to describe the way metaphorical operations within a same text can hold together, notably in terms of thematic coherence. Interestingly, coherence in itself is a complex notion which can be characterised by means of many different types of relationships between discourse constituents; this is the main tenet of approaches to discourse such as Rhetorical Structure Theory (see Mann & Thompson 1988), which postulates that the relationship of justification is one such relationship of coherence. In other words, argumentation is one way of building coherence. Yet, it has also been argued, from a cognitive perspective, that coherence can be reinterpreted in terms of relevance (see Reboul & Moeschler 1996). Under this view, judgements of coherence follow from the perception of relevance between discourse constituents. Therefore, if the multiple occurrences of an extended metaphor throughout a text are not only mutually relevant but also relevant with respect to the core metaphor exploited in the text, addressees will be drawn to perceive the overall extended metaphor as relevant. Owing probably to the literary nature of the corpus analysed therein, the research referred to in what precedes tackles little more than effects of meaning, even if some do point to the argumentative and persuasive dimension metaphors may exploit (e.g. Koller 2005, p. 120). We argue that in particular cases for instance in political discourse extended metaphors are ideally suited to contribute material that can be used for argumentative purposes, this being made possible by the subordination of metaphorical occurrences, throughout the text, to a core metaphor (see section below for a development of this idea). We suggest that extended metaphors can be used to convince their audience of the truth of certain states of affairs, this in turn making them good candidates to implement persuasive strategies. The various occurrences of an extended metaphor in a text can therefore be argumentatively used so as to function as a set of justifications for the metaphor in some sort of confirmational dynamics: every occurrence of an additional aspect of the source domain, to the extent that it is mapped onto the target domain in a plausible (see below) manner, may serve as a confirmation of the overall relevance of the initial metaphorical construal Metaphor and argumentation Metaphor in argumentation theory Literature on the role of metaphor in argumentation is not very extensive (see Santibáñez 2010). As far as modern-day argumentation theory is concerned, Perelman & Olbrechts-Tyteca (2008[1958]) are probably among the researchers who addressed the issue of metaphor in most detail in the 20 th Century. Their account of metaphor within argumentation is formulated along the lines of arguments by analogy, thereby making argumentative metaphors a subtype of arguments establishing the structure of reality: We could not, at this point describe metaphor better than by conceiving of it, at least in what regards argumentation, as a condensed analogy, resulting from the fusion of an element of the phoros with an element of the theme (Perelman & Olbrechts-Tyteca 2008, p. 535). 6 In the New Rhetoric, metaphors are moreover said to play a complex instrumental role as contents which are used to legitimise analogies, inasmuch as they can intervene to accredit the analogy (ibid. p. 536), but also as contents that are derivable on the basis of the analogy itself: (...) oftentimes the author does not hesitate, during his 6 All translations from original texts in French are ours. 7

8 presentation, to make use of metaphors which have been derived from the proposed analogy, thereby habituating the reader to see things as he [the author] shows them to him (ibid.). Although this account is more focused on the argumentative nature of metaphor in terms of argument schemes, it does make room for remarks on the rhetorical advantages of metaphorical discourse, as it also considers that the fusion of the theme and the phoros, which brings their respective domains together, facilitates the realisation of argumentative effects (ibid.). And crucially, we find an idea here that is very relevant to the purpose of this paper and which we will echo in cognitive terms further along, namely the idea that exploiting a mapping between source and target domain at length (i.e., repeatedly) is one way of making the argument stronger in terms of its convincingness: When, through the development of an analogy, we strive to draw conclusions from the phoros that are relevant to the theme, the force of the argument will be more important if, thanks to the fusion of the theme and the phoros, we have previously described at length the phoros in terms of the theme (ibid.). Translated in cognitive linguistic terms, Perelman & Olbrechts-Tyteca may be understood to contend that the more you exploit the metaphor, i.e., the more you instantiate your target domain in terms of your source domain in an argument, the stronger the argument. We will also shortly show how this very idea can receive a cognitive reinterpretation in terms of relevance. Going along the rhetorical path, more recent insights can be found in Plantin s work (2011) on the topic. Following Perelman & Olbrechts-Tyteca, Plantin considers that argumentative metaphors rest on a mechanism of analogy, and that extended metaphors in particular are to be construed as specific forms of structural analogy, whereby the argumentative operation consists in drawing the sceptic s attention to the fact that if the domains are analogous, then their corresponding elements are too (...) (Plantin 2011, p. 120). More interestingly, he mentions that the rhetorical effectiveness of a metaphor is partly determined by the addressee s inferential input in working out metaphorical meaning: From a rhetorical point of view, it [the metaphor] has been appraised as a condensed comparison, the elucidation of which has been entrusted to the audience (ibid., p. 122, our italics). What we think is at stake here is some kind of pleasure or cognitive reward already present in Aristotle s characterisation as noted by Kirby (1997) associated to the full grasping of the metaphorical meaning; in this sense, the addressee s cognitive processing in comprehending the metaphor and its relevance is necessary, even if probably not sufficient, for the metaphor to be rhetorically effective. Another element in Plantin s work is worth mentioning here, as our cognitive account of metaphor as argument will develop along similar lines: it has to do with the status of extended metaphors after they have been exploited at length. Plantin considers that the argumentative force of metaphor is due not only to the fact that, like analogy, it introduces a model of the target situation, but also to the fact that it pushes the analogy over to the point of identification (ibid., p. 123). Extended metaphors, provided they are exploited in a way that is perceived as relevant by the audience, can lead to an identification of source and target domains. If we take the argument further, this could mean that argumentatively exploited metaphors can lead to a demetaphorisation of the construal. In so doing, extended metaphors might be able to provide grounds for a whole new set of further inferences that become, pretty much as Perelman & Olbrechts-Tyteca contend, legitimised or accredited by a metaphor. While Perelman & Olbrechts-Tyteca construe metaphors as taking part in the structure of arguments from analogy, Santibáñez departs from this conception by arguing that arguments 8

9 from metaphor are not necessarily arguments from analogy, mainly on the grounds that similarity is not as important in metaphor as it is in analogy (see Santibáñez 2010, p for a discussion). He then moves on to discussing the role of metaphors in argumentative chains, and in doing so, he focuses on metaphor not in structural terms, but in terms of content and in terms of the argumentative function it might fulfil. His conclusion, dovetailing with the Toulminian perspective, is that metaphors can act as backings. In other words, metaphors have legitimising potential in so far as they can function as major premises in an argument. For instance, if I want to argue that one individual is responsible for the decay of an entire community by uttering Jeffrey is the rotten apple of our community, I may resort to the proverb A rotten apple spoils the barrel : the informational import of the proverb will be selected (and thus deemed relevant) as the major premise linking the standpoint ( Jeffrey is responsible for the decay of the whole community ) and the premise that Jeffrey is metaphorically the rotten apple of the barrel (i.e., the community in which he lives). This is something that is also relevant to our own purposes, to the extent that Santibáñez focuses on the representational as opposed to the formal aspects of metaphors in argumentation. Beyond its relationship with analogy, metaphor is above all a conceptual phenomenon responsible for the generation of particular representations which people can communicate; paying attention to the specificities of these representations becomes relevant if we are to perform a rhetorical analysis of arguments from metaphor. Although some other studies in argumentation revolving around the notion of metaphor (see e.g. Plug & Snoeck Henkemans 2008, Lakoff 2006, Pielenz 1993) are available, for the purposes of this paper we will restrict ourselves to the aspects discussed above. Suffice it to highlight for the time being that metaphors have been said to play a fundamental role in argumentation in terms of the rhetorical possibilities they open up and in terms of the constraints they can impose on the representation of propositions that will be used as premises and conclusions in argumentative chains Extended metaphors as argumentative devices: a cognitive take 7 In section 2.2 we have evoked how extended metaphors are inherently discursive, insofar as they set up a core metaphorical construal and exploit it throughout discourse. Such a discursive structure in our view makes for an argumentative exploitation. The main idea behind our argumentative account of extended metaphors is that they are ideally suited, both discursively and cognitively, to provide an argumentative structure where the initial metaphorical construal becomes a standpoint, while the various instantiations of the metaphor throughout the text function as arguments in support of this standpoint. In political discourse in particular, where speakers constantly try to convey the legitimacy of their claims, this can become an argumentative strategy. Once the metaphor is set up at the beginning of the text, its extension, by mention of several of its properties, may provide evidence to legitimise the initial construal. In order to illustrate this, let us extend Chilton s analysis of Hitler s Mein Kampf (Chilton 2005). In the chapter titled Volk und Rasse, Hitler conceptualises the Jew as a parasite and tries to justify that it makes sense to take on the proposed construal. In order to do this, he 7 We thank an anonymous reviewer for their insightful questions on a first draft of this section, which allowed us to sharpen our account. 9

10 lists some of the properties associated to parasites and observes that these are also typical of the Jews. According to Hitler, just like parasites, Jews i) are not nomads, ii) spread, iii) are thrown out of host nations, iv) seek a new feeding ground when this happens, v) drain their hosts from their resources, etc. Hitler thus attempts to systematically map the conceptual properties we normally associate to parasites onto the different properties, in terms of behaviour, of the Jews. The inherent structure of extended metaphor can thus provide an ideal argumentative ground to give weight to the metaphorical construal. We show in more detail in section 3 how this can be achieved. The cognitive benefits of this rhetorical strategy now need to be spelled out. We postulate here that the more plausible the metaphorical mappings exploited in an extended metaphor are (i.e., the more the addressee perceives that the construal of the target domain in terms of the source domain is justified), the more its overall perceived relevance increases. Our use of the notion of plausibility is grounded on the notion of justification. A metaphorical mapping will thus be deemed plausible if it is justified, i.e., if the content of the particular instances of the metaphor are interpretable in terms of arguments. We do not need to impose restrictions on the type of arguments that may be used to this end for the time being and this will remain an open question. We will nevertheless see in our analysis in section 3 that arguments by example can provide powerful evidence for the justification of a metaphorical construal. To go back to Hitler s example, the strategy would unfold as follows: once the core mapping linking Jews and parasites has been established, the extension of the metaphor could take each property of the parasite one by one and systematically link each of those properties to an actual state of affairs. The accumulation of concrete examples illustrating that the behaviour of Jews matches the different conceptual properties of parasites would then constitute evidence that it does make sense to construe them as parasites. At the end of the text, the hearer is left out with a correspondence matrix pointing to the plausibility of the initial metaphor. In principle, thus, extended metaphors may be used for legitimation purposes. But there is more: sometimes extended metaphors can go beyond the legitimation of the initial metaphorical construal s meaningfulness: we claim that extended metaphors may even be used to provide evidence for the truth of a claim with the provision of an enabling context. Of course, in the case of literary texts, it would make little sense to argue that an extended metaphor is meant to convince the reader of the literal truth of what was initially construed as a metaphor. 8 However, in political discourse the situation might be different, partly because of the genre (political discourse is inherently argumentative and prone to containing standpoints whose truth is justified by the speaker) and partly because the speaker might initially present a metaphor, which, by being extended, ends up considered as a literal statement in order to invite concrete reactions on behalf of the addressees. Quite paradoxically, thus, in light of concrete evidence (in case the construal is extended so as to appear to be literally true for an unusually important number of properties) an extended metaphor may become less and less metaphorical throughout the text. We could then imagine cases in which what justifies the relevance of the extended metaphor is the impression that the metaphor fades away; under these circumstances, extended metaphors might actually become more relevant by failing to meet the expectations of non-literalness raised by their initial metaphorical status. In other words, we hypothesise that 8 The anonymous reviewer we referred to in the previous footnote rightly pointed out that one would never infer from Carl Sandburg s poem ( The fog comes on little cat feet. / It sits looking over harbour and city on silent haunches and then moves on. ) that the fog literally has feet. 10

11 extended metaphors may lead their addressee to eventually abandon the metaphorical construal altogether and this is part of an argumentative strategy. We claim that this is what happens in the corpus we will analyse below (see section 3.3.). In order to specify this claim on cognitive grounds, we will use the notion of relevance to an individual (Sperber & Wilson 1995, p ). The general idea is that utterances might be deemed relevant not only because they contribute to comprehension, but also because they are instrumental to the satisfaction of other cognitive functions, such as the reorganisation of existing knowledge, or the elaboration of rational desires (Sperber & Wilson 1995, p. 266). To these, we could also add cognitive or epistemic goals such as for instance gaining a better representation of how the world is in order to take action on more reliable grounds, which are goals that go beyond the comprehension of the information contained in the stimulus. On this count, we assume that creative and extended metaphors perceived as relevant can also contribute to the long-term stabilisation of a given representation of the world, to the extent that they may be shown as faithful and accurate descriptions. Creative and extended metaphors perceived as relevant in terms of their comprehension can subsequently be relevant for other purposes, including epistemic or argumentative ones. In the case of political discourse, extended metaphors can give rise to the representation of gradual and near-systematic confirmations that it makes sense to connect the source and target domains of the metaphor in the proposed way. One can consequently end up considering that there are solid grounds to conceptualise the target domain in terms of the source domain. Furthermore, numerous relevant instantiations of the mapping, within an enabling context, can have argumentative implications: the more an addressee is led to find confirmations that the proposed construal is applicable, the more its content might be believed to reflect actual states of affairs, which can eventually lead him to consider that the (initially) metaphorical mapping is actually true i.e., literal. In such cases, the extended metaphor could be thought to cognitively function as an argument meant to demetaphorise the metaphor. The epistemic potential of metaphor has already been proposed in the past: it echoes for instance some of the assumptions underlying Chilton s analysis of fragments of Hitler s Mein Kampf (Chilton 2005). In this paper, Chilton defends the use of conceptual metaphor theory and blending theory to explain why ideas are influential, i.e., why and how they propagate and generate stable representations. To this effect, Chilton states that conceptual constructs can become meme-like and infect the mind (under the right social conditions) when they have complex blending potential that recruits fundamental knowledge domains along with the core mechanisms of metaphor. There is a further ingredient that seems to go along with textualised memes of this kind the delivery of some kind of credibility assurance and epistemic warrant. (Chilton 2005, p. 40, our italics). According to Chilton, metaphors spread, at least partly, because their propositional meaning is believable and additionally carries some sort of epistemic warrant, which is precisely what we are concerned with in this paper. The claim we put forth is that recurring metaphorical entailments exploited by extended metaphors contribute to establishing said epistemic warrants, i.e., justifications of the metaphor s content overall adequacy and relevance, which incidentally make for the warrant of truth and relevance (Chilton 2005, p. 41) recipients of political discourse usually expect to find. By overall adequacy and relevance of a metaphor s content, we mean that the propositional content of the metaphor can be perceived by an 11

12 addressee to correspond to a faithful representation. To go back to Chilton s analysis of Hitler s words, the idea is that if we are exposed to a series of examples, illustrations and arguments to support the view that the Jews are parasites, we may end up adopting the representation that the Jews are parasites (whatever this might exactly and contextually mean). The metaphor s contribution to the propagation of an idea is a perlocutionary matter, 9 as it concerns issues that have to do with the effect of the metaphor on people s minds, and, in some cases, also on people s behaviour. Research in cognitive linguistics on the rhetorical potential of metaphor (e.g. Chilton 2005, Lakoff 2006, Charteris-Black 2006) has shown that in principle metaphors and argumentative utterances are able to generate similar effects. The following section will be devoted to examining the implications of this claim should we generalise it to postulate the argumentative potential of metaphorical expressions, and in particular of extended metaphors Extended metaphors as conceptual argumentations: epistemic issues Extended metaphors require the addressee to assess different aspects of the same metaphorical mapping within the same text and can in principle be exploited to strengthen the perceived relevance of the construal in a specific way, i.e., by achieving a de-metaphorisation of the initial metaphor. The relationship between each occurrence of a particular conceptual mapping of properties between source and target domain and the extended metaphor as a whole can thus be seen as one of argumentative confirmation of relevance with epistemic implications; if every mapped conceptual property is found relevant, chances are that this will also count as a justification of the overall extended metaphor s relevance (in terms of conceptualising the target domain through the source domain), and, as we shall argue, as an attempt to convince the addressee, in light of a complex correspondence matrix, that the proposed construal can in fact be taken as literal. In other words, by inviting the addressee to process in depth and at length different aspects of the extended metaphor, the speaker is weaving an argumentative path in which the accuracy, legitimacy and/or relevance of the metaphorical mapping is the standpoint, while the different instantiations of conceptual mappings for different properties count as arguments in support of said standpoint. Before we turn to the analysis of the data, let us further elaborate this idea of metaphor as argument. The assumption that extended metaphors can be conceptualised as complex argumentations bears at least two implications for their study within a discourse analytical framework. First, it implies that extended metaphors, as any argumentative device, can be used to convince: in this case, this perlocutionary goal translates into attempting to convince the addressee that there are solid grounds to legitimate the truth of what is metaphorically presented at first, and, ultimately, to convince the addressee that there are grounds that come from facts and states of affairs which will allow us to believe the truth of the conceptual construal and accordingly adopt adequate courses of action. 10 Second, it means that metaphors, just like arguments, may bear a rhetorical 9 Recall that Austin s first example of perlocution in the William James Lectures is the act of persuading (cf. Austin 1962, p. 101). 10 We will propose in our analysis that this is what happens in Freysinger s pamphlet: the initial metaphor of the USA as an empire is gradually weakened as a metaphor and strengthened as a propositional content throughout the text in order to lead the addressee to the conclusion that there are necessary actions to undertake in order to protect Switzerland from the threat. 12

13 function, 11 and that as such they are exploitable as rhetorical strategies, otherwise commonly observed in argumentative practices. One subset of rhetorical strategies of particular interest to us here lies in the various ways in which a speaker can positively affect her ethos, which can roughly be defined as the image the speaker wishes to convey through her words, particularly in terms of trustworthiness, consistency, competence and benevolence. The link between rhetoric and pragmatics has already been highlighted in terms of the relationship between rhetorical figures and conversational aspects of communication, notably with respect to politeness and face-management issues (see e.g. Kerbrat-Orecchioni 1994, 2002). Following and extending this line of argument, we believe that a pragmatic analysis of metaphor can also incorporate a rhetorical dimension that sheds light on the effectiveness of extended metaphors with respect to belief fixation, via the examination of the effects that the use of an extended metaphor that turns out to be perceived as relevant can have on perceived speaker competence. We therefore postulate that, just like well-thought argumentative demonstrations, well-thought extended metaphors which are recognised as such can positively influence the speaker s perceived image. Rhetoric considers since Aristotle that the effectiveness of arguments depends on at least three factors: they may convince by virtue of their rational and logical qualities, their ability to resonate with their audience s beliefs, desires and emotional states, but also by virtue of the speaker s charisma, perceived trustworthiness, benevolence and competence. 12 We postulate that some effects of extended metaphors can be interpreted as rhetorical, notably in what regards speaker ethos. The argument goes like this: if the various occurrences of an extended metaphor in a text achieve overall relevance in his cognitive environment, then the addressee will be led to entertain a rich and sophisticated representation that he takes to correspond to a state of affairs the speaker intended to communicate. To the extent that he recognises this richness and sophistication, he might conclude that only a knowledgeable and skilful speaker can be responsible for such a dense and presumably accurate construal. In other words, the hearer can tentatively be led to postulate the speaker s competence, for he himself (the hearer) would not have been able to come up with such a level of conceptual complexity. If the extended metaphor presents relevant mappings down the line in every occurrence of the metaphor, chances are that in the end, the hearer will be left with the impression that the speaker perfectly knows how to handle her story. Specifically, recognising that each occurrence of the metaphor is relevant by being led to infer that every property instantiated echoes an actual state of affairs might be seen as analogous to recognising the cogency and validity that accompanies a solid argumentative chain. The abstract form of a rhetorically successful extended metaphor can this way be conceptually compared to the abstract form of a rhetorically successful complex argument: in such an argument the standpoint S would be presented as true or acceptable by providing arguments X, Y, Z in support of S; similarly, in a rhetorically successful extended metaphor, the primary metaphorical mapping is presented as highly relevant and, by successively advancing specific aspects of said mapping that act as arguments in its support, it might come to be believed as referring to an actual state of affairs. Each correspondence thus functions as an argument supporting the 11 We are here concerned with the argumentative dimension of rhetoric not so much with its stylistic dimension and with the role metaphors play in convincing or persuading the addressee. 12 These are obviously the respective components of the traditional rhetorical triangle composed of logos, pathos, and ethos. 13

14 conceptual mapping s adequacy to the world, thereby de-metaphorising the initial construal. The fact that the addressee recognises, based on a rich matrix of correspondences, that the extended metaphor is relevant will arguably lead him to consider that only a competent and smart speaker (therefore perceived as someone clever, who has spent significant time and effort reflecting upon the issue, and by extension someone we can trust) would have been able to conceptualise the correspondences in such a way. Along these lines, if we consider that extended metaphors can have a self-argumentative function of legitimation, it does make sense to describe them as argumentative devices. They draw on linguistic and cognitive mechanisms that are at play when people assess not only speaker meaning, but also epistemic aspects of their content, i.e., clues as to whether the information they convey should be accepted or not. Recent work in cognitive and evolutionary anthropology (cf. Sperber et al. 2010) has postulated that the selection of communication as a stable property of human life through evolution supposes that its benefits have outweighed its disadvantages in the long run, and thus that the human species has found a way of dealing with cheaters and deceivers while still preserving communication. For Sperber et al., this means that we have presumably developed some abilities to spot cheaters and to assess the reliability of communicated material, some kind of suite of cognitive mechanisms for epistemic vigilance, targeted at the risk of being misinformed by others (Sperber et al. 2010: 359). From this perspective, epistemic vigilance is defined as a cognitive ability which is intimately linked to the possibility of communication, in that it appears to be a necessary condition for communication to stabilise. Sperber et al. indeed note that a disposition to be vigilant is likely to have evolved biologically alongside the ability to communicate in the way that humans do (Sperber et al. 2010: ). Epistemic vigilance mechanisms are globally believed to be responsible for the assessment of i) the source of the message and ii) the message s content. As far as the source is concerned, addressees will typically try to assess speaker competence (one s possessing genuinely reliable information) and benevolence (i.e., the intention of sharing that genuine information) and try to look for evidence of the speaker s reliability and overall trustworthiness in order not to be misled or cheated. Assessment of the source requires a metarepresentational ability, since it supposes the necessity of being able to make assumptions about a speaker s intentions. As far as message assessment is concerned, the account of epistemic vigilance basically relies on the relevance-theoretic framework: the search for a relevant interpretation, which is part and parcel of the comprehension process, automatically involves the making of inferences which may turn up inconsistencies or incoherences relevant to epistemic assessment (Sperber et al. 2010: 376). Scanning the source of the message in order to establish its trustworthiness typically involves assessing whether the speaker is competent. This is where the link with the rhetorical potential of extended metaphors used in argumentative contexts comes in: if a speaker is able to work out an extended metaphor in such a way that what was initially taken as a metaphor ends up being believed as literal, chances are that the addressee will consider that the speaker is competent, and therefore trustworthy, which is a condition that maximises the chances of belief fixation we believe what the speaker says because we find her reliable. Following this line of argument, we hypothesise accordingly that rhetorically successful extended metaphors match the minimal conditions required to go past our epistemic vigilance monitoring; this, we contend, may in turn be achieved via the implications noted above on speaker ethos. 14